Currently, an object-based audio encoding technology that may efficiently compress an audio object signal is the focus of attention. A quantization/dequantization scheme of a parameter for supporting an arbitrary downmix signal of an existing Moving Picture Experts Group (MPEG) Surround technology may extract a Channel Level Difference (CLD) parameter between an arbitrary downmix signal and a downmix signal of an encoder. Also, the quantization/dequantization scheme may perform quantization/dequantization using a CLD quantization table symmetrically designed based on 0 dB in an MPEG Surround scheme.
A mastering downmix signal may be generated when a plurality of instruments/tracks are mixed as a stereo signal, are amplified to have a maximum dynamic range that a Compact Disc (CD) may represent, and are converted by an equalizer, and the like. Accordingly, a mastering downmix signal may be different from a stereo mixing signal.
When an arbitrary downmix processing technology of an MPEG Surround scheme is applied to a multi-object audio encoder to support a mastering downmix signal, a CLD between a downmix signal and a mastering downmix signal may be asymmetrically extracted due to a downmix gain of each object. Here, the CLD may be obtained by multiplying each of the objects with the downmix gain. Accordingly, only one side of an existing CLD quantization table may be used, and thus a quantization error occurring during a quantization/dequantization of a CLD parameter may be significant.
Accordingly, a method of efficiently encoding/decoding an audio object is required.